一种小波神经网络结构及其学习算法研究

A Novel Structure of Wavelet Neural Network and Its Learning Algorithm

基于仿射小波神经网络的逼近原理和结构设计问题 ,本文提出了一种新型小波神经网络结构 ,研究了其结构化设计方法和相应的学习算法 ,优化组合小波基元激励函数 ,实现了小波神经网络权系数的二次学习 ,避免了“维数灾”问题 ,改善了网络学习特性。计算机仿真结果表明 ,研究的小波神经网络结构及其学习算法简单有效 。

Based on the function approximation theory and the structure design of wavelet neural network, a novel structure of wavelet neural network is proposed in this paper. Its structure design method and corresponding learning algorithm are developed. The second learning of weight coefficients can be realized, the 'curse of dimensionality' is avoided and the network learning algorithm is also improved. The simulation result shows the efficiency and the excellent performance of this novel wavelet neural network and its learning algorithm.